With the advent of creating intelligent indoor spaces across diverse domains such asdomestic, retail, industries, and healthcare, there has been an intensified demand foradvanced indoor localization systems, especially with the integration of autonomousmachinery and robots into daily life. This thesis aims to surpass the limitations of currentRF- and Optical-based localization systems by developing more accurate, reliable, energy-efficient, and cost-effective indoor localization systems. These systems are envisioned forservices like navigation in autonomous vehicles and asset tracking in industrial settings. Thethesis categorizes localization systems into two types: autonomous systems for entities likewarehouse robots, and tracking systems for monitoring indoor assets.Initially, the thesis introduces a Bluetooth Low Energy (BLE)-based autonomous localization system, BLoB. This system employs synchronous transmissions for enhanced reliabilityand energy efficiency, achieving sub-meter-level accuracy with single-antenna BLE devices.BLoB capitalizes on a unique beating effect observed in synchronous transmissions, characterized by a sinusoidal pattern of constructive and destructive interference. Utilizing theconstant tone extension feature of the BLE 5.1 standard, BLoB enables multiple anchors totransmit synchronously, with mobile tags capturing the resulting signal. The identification ofpeaks in this superimposed signal, combined with signal strength data, empowers BLoB toattain sub-meter accuracy in positioning. The efficacy of BLoB is validated in various settings,including offices and sports halls, proving its robustness in challenging indoor scenarios.To further refine BLE localization to centimeter or decimeter levels, the thesis proposes theintegration of BLE with precise optical-based localization methods. It introduces BLELight, ahybrid autonomous localization system that merges the capabilities of BLE and Visible LightPositioning (VLP) technology through neural network-driven data fusion. The deep neuralnetwork model, trained through an incremental learning approach, showcases a substantialimprovement in localization accuracy, achieving decimeter-level accuracy.However, prior to developing this hybrid solution with VLP, the thesis presents an innovative VLP system named HueSense . The system utilizes existing LED lighting infrastructure as transmitters for location beacons. This innovative system requires no modifications toinstalled LEDs, instead analyzing their intrinsic characteristics for differentiation and locationmapping. HueSense employs low-power, off-the-shelf hue sensors to realize this functionality.A prototype using three hue sensors, tested under different lighting environments, demonstrates the system’s practicality and accuracy, either as a standalone solution or in enhancing BLE-based localization. BLELight, by integrating features of both BLoB and HueSense , evolvesinto a comprehensive hybrid autonomous localization system. This system adeptly tackles real-world challenges such as interference from external ambient light, physical obstructions, and shadows, issues that typically limit VLP efficacy, while also addressing the inherent accuracyconstraints of BLE technology.Further, the thesis explores a tracking system that leverages the new direction-findingtechniques of the BLE 5.1 standard, augmented with mmWave radar measurements. This integration, trained jointly in a deep neural network model, addresses the inherent accuracy limitations of BLE. Two variations, BmmW-LITE and BmmW-LITE+ , are evaluated. These systemsare optimized for single-antenna BLE devices, eliminating the need for bulky multi-antennaarrays and presenting a more compact, cost-effective solution for IoT devices. BmmW-LITE+extends BmmW-LITE by incorporating semantic capabilities at the edge device, facilitatingdata transfer from the edge to the cloud, and optimizing bandwidth, power, and memory usage.All systems, tested experimentally, demonstrate decimeter-level localization accuracy and canbe easily incorporated into solutions for deployment in real-world indoor environments.
(2024). Indoor Localization Systems with Bluetooth and Light: Design and Implementation.
Indoor Localization Systems with Bluetooth and Light: Design and Implementation
SINGH, Jagdeep
2024-03-21
Abstract
With the advent of creating intelligent indoor spaces across diverse domains such asdomestic, retail, industries, and healthcare, there has been an intensified demand foradvanced indoor localization systems, especially with the integration of autonomousmachinery and robots into daily life. This thesis aims to surpass the limitations of currentRF- and Optical-based localization systems by developing more accurate, reliable, energy-efficient, and cost-effective indoor localization systems. These systems are envisioned forservices like navigation in autonomous vehicles and asset tracking in industrial settings. Thethesis categorizes localization systems into two types: autonomous systems for entities likewarehouse robots, and tracking systems for monitoring indoor assets.Initially, the thesis introduces a Bluetooth Low Energy (BLE)-based autonomous localization system, BLoB. This system employs synchronous transmissions for enhanced reliabilityand energy efficiency, achieving sub-meter-level accuracy with single-antenna BLE devices.BLoB capitalizes on a unique beating effect observed in synchronous transmissions, characterized by a sinusoidal pattern of constructive and destructive interference. Utilizing theconstant tone extension feature of the BLE 5.1 standard, BLoB enables multiple anchors totransmit synchronously, with mobile tags capturing the resulting signal. The identification ofpeaks in this superimposed signal, combined with signal strength data, empowers BLoB toattain sub-meter accuracy in positioning. The efficacy of BLoB is validated in various settings,including offices and sports halls, proving its robustness in challenging indoor scenarios.To further refine BLE localization to centimeter or decimeter levels, the thesis proposes theintegration of BLE with precise optical-based localization methods. It introduces BLELight, ahybrid autonomous localization system that merges the capabilities of BLE and Visible LightPositioning (VLP) technology through neural network-driven data fusion. The deep neuralnetwork model, trained through an incremental learning approach, showcases a substantialimprovement in localization accuracy, achieving decimeter-level accuracy.However, prior to developing this hybrid solution with VLP, the thesis presents an innovative VLP system named HueSense . The system utilizes existing LED lighting infrastructure as transmitters for location beacons. This innovative system requires no modifications toinstalled LEDs, instead analyzing their intrinsic characteristics for differentiation and locationmapping. HueSense employs low-power, off-the-shelf hue sensors to realize this functionality.A prototype using three hue sensors, tested under different lighting environments, demonstrates the system’s practicality and accuracy, either as a standalone solution or in enhancing BLE-based localization. BLELight, by integrating features of both BLoB and HueSense , evolvesinto a comprehensive hybrid autonomous localization system. This system adeptly tackles real-world challenges such as interference from external ambient light, physical obstructions, and shadows, issues that typically limit VLP efficacy, while also addressing the inherent accuracyconstraints of BLE technology.Further, the thesis explores a tracking system that leverages the new direction-findingtechniques of the BLE 5.1 standard, augmented with mmWave radar measurements. This integration, trained jointly in a deep neural network model, addresses the inherent accuracy limitations of BLE. Two variations, BmmW-LITE and BmmW-LITE+ , are evaluated. These systemsare optimized for single-antenna BLE devices, eliminating the need for bulky multi-antennaarrays and presenting a more compact, cost-effective solution for IoT devices. BmmW-LITE+extends BmmW-LITE by incorporating semantic capabilities at the edge device, facilitatingdata transfer from the edge to the cloud, and optimizing bandwidth, power, and memory usage.All systems, tested experimentally, demonstrate decimeter-level localization accuracy and canbe easily incorporated into solutions for deployment in real-world indoor environments.File | Dimensione | Formato | |
---|---|---|---|
Jagdeep_Doctoral_Thesis_2024_Localization-Systems.pdf
accesso aperto
Descrizione: PhD Thesis
Tipologia:
Tesi di dottorato
Dimensione
39.6 MB
Formato
Adobe PDF
|
39.6 MB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.